add doc ori cls model (#2018)

* add doc ori cls model

* add doc ori cls model

* add doc ori cls model

* fix class num

README.md

@@ -59,11 +59,14 @@ PaddleX 3.0 覆盖了 16 条产业级模型产线，其中 9 条基础产线可
     <th>具体模型</th>
   </tr>
   <tr>
-    <td>基础产线</td>
-    <td>通用图像分类</td>
+    <td rowspan="2">基础产线</td>
+    <td rowspan="2">通用图像分类</td>
     <td>图像分类</td>
     <td>CLIP_vit_base_patch16_224<br/>CLIP_vit_large_patch14_224<details>
     <summary><b>more</b></summary><br/>ConvNeXt_tiny<br/>ConvNeXt_small<br/>ConvNeXt_base_224<br/>ConvNeXt_base_384<br/>ConvNeXt_large_224<br/>ConvNeXt_large_384<br/>MobileNetV1_x0_25<br/>MobileNetV1_x0_5<br/>MobileNetV1_x0_75<br/>MobileNetV1_x1_0<br/>MobileNetV2_x0_25<br/>MobileNetV2_x0_5<br/>MobileNetV2_x1_0<br/>MobileNetV2_x1_5<br/>MobileNetV2_x2_0<br/>MobileNetV3_large_x0_35<br/>MobileNetV3_large_x0_5<br/>MobileNetV3_large_x0_75<br/>MobileNetV3_large_x1_0<br/>MobileNetV3_large_x1_25<br/>MobileNetV3_small_x0_35<br/>MobileNetV3_small_x0_5<br/>MobileNetV3_small_x0_75<br/>MobileNetV3_small_x1_0<br/>MobileNetV3_small_x1_25<br/>PP-HGNet_tiny<br/>PP-HGNet_small<br/>PP-HGNet_base<br/>PP-HGNetV2-B0<br/>PP-HGNetV2-B1<br/>PP-HGNetV2-B2<br/>PP-HGNetV2-B3<br/>PP-HGNetV2-B4<br/>PP-HGNetV2-B5<br/>PP-HGNetV2-B6<br/>PP-LCNet_x0_25<br/>PP-LCNet_x0_35<br/>PP-LCNet_x0_5<br/>PP-LCNet_x0_75<br/>PP-LCNet_x1_0<br/>PP-LCNet_x1_5<br/>PP-LCNet_x2_0<br/>PP-LCNet_x2_5<br/>PP-LCNetV2_small<br/>PP-LCNetV2_base<br/>PP-LCNetV2_large<br/>ResNet18<br/>ResNet18_vd<br/>ResNet34<br/>ResNet34_vd<br/>ResNet50<br/>ResNet50_vd<br/>ResNet101<br/>ResNet101_vd<br/>ResNet152<br/>ResNet152_vd<br/>ResNet200_vd<br/>SwinTransformer_tiny_patch4_window7_224<br/>SwinTransformer_small_patch4_window7_224<br/>SwinTransformer_base_patch4_window7_224<br/>SwinTransformer_base_patch4_window12_384<br/>SwinTransformer_large_patch4_window7_224<br/>SwinTransformer_large_patch4_window12_384</details></td>
+  <tr>
+    <td>文档图像方向分类</td>
+    <td>PP-LCNet_x1_0_doc_ori</td>
   </tr>
   <tr>
     <td>基础产线</td>

docs/tutorials/models/support_model_list.md

@@ -107,6 +107,11 @@
 | PP-HGNetV2-B6_ML | [PP-HGNetV2-B6_ML.yaml](../../../paddlex/configs/multilabel_classification/PP-HGNetV2-B6_ML.yaml)|
 | CLIP_vit_base_patch16_448_ML | [CLIP_vit_base_patch16_448_ML.yaml](../../../paddlex/configs/multilabel_classification/CLIP_vit_base_patch16_448_ML.yaml)|
 
+### 12. 文档图像方向分类
+| 模型名称 | config |
+| :--- | :---: |
+| PP-LCNet_x1_0_doc_ori | [PP-LCNet_x1_0_doc_ori.yaml](../../../paddlex/configs/doc_text_orientation/PP-LCNet_x1_0_doc_ori.yaml)|
+
 ## 二、目标检测
 ### 1. PP-YOLOE_plus 系列
 | 模型名称 | config |

paddlex/configs/doc_text_orientation/PP-LCNet_x1_0_doc_ori.yaml

@@ -0,0 +1,41 @@
+Global:
+  model: PP-LCNet_x1_0_doc_ori
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/cls/cls_flowers_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 102
+  epochs_iters: 50
+  batch_size: 16
+  learning_rate: 0.08
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+  save_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Export:
+  weight_path: https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/PPLCNet_x1_0_docori_pretrained.pdparams
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/img_rot180_demo.jpg"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/modules/base/predictor/utils/official_models.py

@@ -34,6 +34,7 @@ OFFICIAL_MODELS = {
     "PP-LCNet_x0_5": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0b1/PP-LCNet_x0_5_infer.tar",
     "PP-LCNet_x0_75": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0b1/PP-LCNet_x0_75_infer.tar",
     "PP-LCNet_x1_0": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0b1/PP-LCNet_x1_0_infer.tar",
+    "PP-LCNet_x1_0_doc_ori": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0b1/PP-LCNet_x1_0_doc_ori_infer.tar",
     "PP-LCNet_x1_5": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0b1/PP-LCNet_x1_5_infer.tar",
     "PP-LCNet_x2_5": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0b1/PP-LCNet_x2_5_infer.tar",
     "PP-LCNet_x2_0": "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0b1/PP-LCNet_x2_0_infer.tar",

paddlex/modules/image_classification/model_list.py

@@ -55,6 +55,7 @@ MODELS = [
     "PP-LCNet_x0_5",
     "PP-LCNet_x0_75",
     "PP-LCNet_x1_0",
+    "PP-LCNet_x1_0_doc_ori",
     "PP-LCNet_x1_5",
     "PP-LCNet_x2_0",
     "PP-LCNet_x2_5",

paddlex/repo_apis/PaddleClas_api/cls/register.py

@@ -171,6 +171,15 @@ register_model_info(
 
 register_model_info(
     {
+        "model_name": "PP-LCNet_x1_0_doc_ori",
+        "suite": "Cls",
+        "config_path": osp.join(PDX_CONFIG_DIR, "PP-LCNet_x1_0_doc_ori.yaml"),
+        "supported_apis": ["train", "evaluate", "predict", "export"],
+    }
+)
+
+register_model_info(
+    {
         "model_name": "PP-LCNet_x1_5",
         "suite": "Cls",
         "config_path": osp.join(PDX_CONFIG_DIR, "PP-LCNet_x1_5.yaml"),

paddlex/repo_apis/PaddleClas_api/configs/PP-LCNet_x1_0_doc_ori.yaml

@@ -0,0 +1,153 @@
+# global configs
+Global:
+  checkpoints: null
+  pretrained_model: null
+  output_dir: ./output/
+  device: gpu
+  save_interval: 10
+  eval_during_train: True
+  eval_interval: 10
+  epochs: 60
+  print_batch_step: 10
+  use_visualdl: False
+  # used for static mode and model export
+  image_shape: [3, 224, 224]
+  save_inference_dir: ./inference
+
+# mixed precision
+AMP:
+  use_amp: False
+  use_fp16_test: False
+  scale_loss: 128.0
+  use_dynamic_loss_scaling: True
+  use_promote: False
+  # O1: mixed fp16, O2: pure fp16
+  level: O1
+
+# model architecture
+Arch:
+  name: PPLCNet_x1_0
+  pretrained: True
+  class_num: 4
+  use_last_conv: False
+
+# loss function config for traing/eval process
+Loss:
+  Train:
+    - CELoss:
+        weight: 1.0
+        epsilon: 0.1
+  Eval:
+    - CELoss:
+        weight: 1.0
+
+
+Optimizer:
+  name: Momentum
+  momentum: 0.9
+  lr:
+    name: Cosine
+    learning_rate: 0.14
+    warmup_epoch: 5
+  regularizer:
+    name: 'L2'
+    coeff: 0.00003
+
+
+# data loader for train and eval
+DataLoader:
+  Train:
+    dataset:
+      name: CustomLabelDataset
+      image_root: ./dataset/OrientationDataset/
+      cls_label_path: ./dataset/OrientationDataset/train_list.txt
+      transform_ops:
+        - DecodeImage:
+            to_rgb: True
+            channel_first: False
+        - RandCropImage:
+            size: 224
+        - TimmAutoAugment:
+            prob: 0.0
+            config_str: rand-m9-mstd0.5-inc1
+            interpolation: bicubic
+            img_size: 224
+        - NormalizeImage:
+            scale: 1.0/255.0
+            mean: [0.485, 0.456, 0.406]
+            std: [0.229, 0.224, 0.225]
+            order: ''
+        - RandomErasing:
+            EPSILON: 0.0
+            sl: 0.02
+            sh: 1.0/3.0
+            r1: 0.3
+            attempt: 10
+            use_log_aspect: True
+            mode: pixel
+        - RandomRot90:
+    sampler:
+      name: DistributedBatchSampler
+      batch_size: 256
+      drop_last: False
+      shuffle: True
+    loader:
+      num_workers: 12
+      use_shared_memory: True
+
+  Eval:
+    dataset: 
+      name: ImageNetDataset
+      image_root: ./dataset/OrientationDataset/
+      cls_label_path: ./dataset/OrientationDataset/val_list.txt
+      transform_ops:
+        - DecodeImage:
+            to_rgb: True
+            channel_first: False
+        - ResizeImage:
+            resize_short: 256
+        - CropImage:
+            size: 224
+        - NormalizeImage:
+            scale: 1.0/255.0
+            mean: [0.485, 0.456, 0.406]
+            std: [0.229, 0.224, 0.225]
+            order: ''
+    sampler:
+      name: DistributedBatchSampler
+      batch_size: 64
+      drop_last: False
+      shuffle: False
+    loader:
+      num_workers: 4
+      use_shared_memory: True
+
+Infer:
+  infer_imgs: ./test_img/
+  batch_size: 1
+  transforms:
+    - DecodeImage:
+        to_rgb: True
+        channel_first: False
+    - ResizeImage:
+        resize_short: 256
+    - CropImage:
+        size: 224
+    - NormalizeImage:
+        scale: 1.0/255.0
+        mean: [0.485, 0.456, 0.406]
+        std: [0.229, 0.224, 0.225]
+        order: ''
+    - ToCHWImage:
+  PostProcess:
+    name: Topk
+    topk: 1
+    class_id_map_file: ppcls/utils/PULC_label_list/text_image_orientation_label_list.txt
+
+Metric:
+  Train:
+    - TopkAcc:
+        topk: [1]
+  Eval:
+    - TopkAcc:
+        topk: [1]